The use cases are different for me, personally. There are some minor (on the surface), but major (depending on how you use the software) differences between them out of the box:

• Logseq is focused on daily journal pages and pages with lots of linkages, the idea being it keeps you on focused on tasks without feeling like you are spending too much time tinkering, organizing, or like you have to build out a perfect system right away. I like it for work because it lets me work "in my notes" rather than needing to work "on my notes" if that makes sense. I'm aware of similar Obsidian setups, but having it work out of the box in Logseq means many of the other design choices in the program are made with this workflow in mind.
• Every note is structured as a tree with hierarchical items (blocks) nested within one another. This means every block has a reference, so it's easy to create and maintain links to different pages. Obsidian does support block references out of the box, but you have to insert them for every line you want to reference because it isn't set up to enforce a tree structure by default.
• Tags and note links are interchangeable. I would actually say this is the main point against vanilla Logseq compared to Obsidian. There are plugins you can use to give tags different behavior to links though.
• The community seems smaller and there aren't as many plugins. Many plugins don't seem to be well supported or maintained, but they are usually pretty focused on solving one particular problem.

I use Logseq for work where being able to reference blocks is more useful (especially for task management), and Obsidian for personal projects where I feel a more free form PKM with customization options is nicer.

Idk, I've never tried Obsidian so it's hard for me to compare, essentially it's a giant bullet-point database. You can add links or tags (they act the same) that act as references to pages, and opening those pages shows you all the references.

The idea is to write without thinking too much about it, and maybe going over it after you wrote to add links if you think it's necessary. Personally, I wish I could join together OneNote (yes I know) and Logseq, because I like reasoning spatially and Logseq's Whiteboards feature doesn't click for me, but it's good for writing down stuff in meetings and so.

Thank you for your answer.
My problem is, that I would like to keep my (Docker) services like Silverbullet.md in my local network only and reach them exterally by Wireguard VPN.
Before https I reached my services with <ip>:<port>, which worked fine with VPN. As mentioned here, there are now only a few options to get https without open everything to the world:

• Set my local servers ip as localhost by ssh ( ssh -N -L 3002:localhost:3000 user@someip) which works, but needs to be added for every client and every access.

• Add a hosts file entry (e.g. silverbullet.lan 192.168.1.123) on every client to access the local server and add a Caddyfile entry with a selfsigned certificate.

• Put a local DNS server like dnsmasq or bind9 in my network and add local domains there (like silverbullet.lan) and use Caddy again for SSL. That's the point where I haven't figured out how to get it running over Wireguard VPN.

I don't want to rely on external services like Tailscale.